[TS] Generating through OpenAI SDK directly

[louis030195]

Describe the feature or potential improvement

import { Langfuse } from "langfuse";
 
const langfuse = new Langfuse();
 
// Example generation creation
const generation = trace.generation({
  name: "chat-completion",
  model: "gpt-3.5-turbo",
  modelParameters: {
    temperature: 0.9,
    maxTokens: 2000,
  },
  input: messages,
});
 
// Application code
const chatCompletion = await llm.respond(prompt);
 
// End generation - sets endTime
generation.end({
  output: chatCompletion,
});

to:

const langfuse = new Langfuse();
 const openai = new OpenAI();


const openai = await openai.completion(prompt);

Additional information

No response

[marcklingen]

Agree, the integration for OpenAI TS could be way better. Adding your +1 to #1065

Related, a more generalized wrapper to automatically instrument functions would be awesome for TS. Currently we are working on decorators for Python (#1009) and this could be a great next step.

[hassiebp]

@louis030195 We've just released the JS integration for the OpenAI SDK in the Langfuse JS SDK 3.6.0 🎉 Feel free to try it out and let us know your feedback 🙏🏾

Link to release
Link to docs